An Adaptive Voltage-Sensor-Based MPPT for ... - IEEE Xplore

IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 62, NO. 12, DECEMBER 2015

7609

An Adaptive Voltage-Sensor-Based MPPT for Photovoltaic Systems With SEPIC Converter Including Steady-State and Drift Analysis Muralidhar Killi and Susovon Samanta, Member, IEEE

Abstract—An adaptive voltage-sensor-based maximum power point tracking algorithm employing a variable scaling factor for a single-ended primary-inductance converter is presented. In this method, only a voltage divider circuit is used to sense the photovoltaic (PV) panel voltage. This method can effectively improve both transient and steadystate performance by varying the scaling factor as compared with the fixed step size and adaptive step size with fixed scaling factor. For sudden change in solar insolation or in start-up, this method leads to faster tracking, whereas in steady state, it leads to lower oscillations around maximum power point. The steady-state behavior and drift phenomena are also addressed in this paper to determine the tracking efficiency. The duty cycle is generated directly without using any proportional–integral control loop to simplify the control circuit. MATLAB/Simulink is used for simulation studies, and a microcontroller is used as a digital platform to implement the proposed algorithm for experimental validation. The proposed system is implemented and tested successfully on a PV panel in the laboratory. Index Terms—Adaptive, drift phenomena, maximum power point tracking (MPPT), photovoltaic (PV), single-ended primary-inductance converter (SEPIC), voltage sensor.

I. I NTRODUCTION

P

HOTOVOLTAIC (PV) power generation is evolving as one of the most prominent renewable energy sources because of its merits such as ecofriendly nature, less maintenance, and no noise. The fundamental component of PV system is a PV cell. Series connection of PV cells forms modules, and series and parallel connection of modules forms arrays. The characteristics of a PV module will vary with solar insolation and atmospheric temperature [1]. The efficiency of the PV system mainly depends on the operating point on the characteristic curve of the PV module. The point at which available maximum power can be extracted from the PV module is called maximum power point (MPP). So far, a large number of maximum power point tracking (MPPT) techniques have been developed [2]–[28] to increase the efficiency of the PV system.

Manuscript received April 21, 2014; revised August 6, 2014, January 5, 2015, and March 13, 2015; accepted May 26, 2015. Date of publication July 20, 2015; date of current version November 6, 2015. The authors are with the Department of Electrical Engineering, National Institute of Technology, Rourkela 769 008, India (e-mail: [email protected]). Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/TIE.2015.2458298

MPPT algorithms such as fractional open-circuit voltage [2], fractional short-circuit current [3], hill climbing [4], perturb and observe (P&O) [5], [6], incremental conductance (IncCond) [7]–[9], incremental resistance [10], ripple correlation control [11], fuzzy logic [12], neural network [13], particle swarm optimization [14], and sliding mode [15], [16] control techniques have been developed to extract the maximum power from the PV arrays. Among the various MPPT techniques, fractional open-circuit voltage and short-circuit current techniques provide a simple and effective way to extract maximum power, but they require periodical measurement of open-circuit voltage or short-circuit current for reference, causing more power loss. Both P&O and hill climbing methods are extensively practiced methods because of their increased efficiency and ease of implementation [17]. However, the sudden changes in atmospheric conditions cause these P&O-like algorithms to drift away from MPP [18]–[20]. According to the literature, the problem of drift is addressed and solved by using IncCond technique [21]. However, the present studies observed that IncCond method also suffers from drift [22]. Other existing techniques show improved performance using fuzzy logic, neural network, optimization algorithm, and sliding mode control, but they are not commonly used due to their complexity and need of expensive digital processor [17]. Overview of all the MPPT techniques recently published is thoroughly discussed in [17], [23], and [24]. The conventional MPPT methods are usually implemented with a fixed perturbation step size determined by the tradeoff between efficiency and tracking speed requirements [7], [25]. Variable-step-size MPPT methods are presented in [4] and [10] to reduce the tracking time and to improve the steady-state performance. The step size is defined as a function of either the derivative of power to voltage (dP/dV ) [7] or the derivative of power to duty cycle (dP/dD) [4]. The adaptive MPPT algorithms immensely increase the efficiency of the system by reducing the tracking time and power loss in steady state [26]. Among the various MPPT techniques, P&O and IncCond are the most widely used techniques. To implement the P&O and IncCond methods, both voltage and current sensors are required. In general, current sensing is done by using a shunt resistor in differential amplifier configurations, but power losses will occur in current conducting path, and the bandwidth is limited by the amplifier. Hall-effect current sensors can provide an alternative option with low loss and good accuracy, but at a higher price; moreover, they are inherently noisy in nature.

0278-0046 © 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

7610


Fig. 1. Block diagram of a PV system with MPPT control.

Thus, an MPPT method with only voltage sensing is more efficient in terms of reduced power loss and low cost. Voltagesensor-based MPPT technique with fixed step size has been developed and is validated for an interleaved dual boost converter in [27]. Later, an adaptive voltage-sensor-based MPPT with a constant start-up scaling factor has been developed by considering dP/dD as an objective function, where P is the power of the PV module, and D is the duty cycle of the converter [28]. A variable scaling factor is applied in the proposed voltage-based adaptive MPPT technique to obtain fast tracking response and reduced steady-state oscillations. Steady-state behavior and drift phenomena are the main concern of any MPPT algorithm to determine the tracking efficiency. These analyses for most popular MPPT methods such as P&O and IncCond methods well exist in the literature [19], [20], [22], but there is no existence of literature analyzing steady-state behavior and drift phenomena for voltage-sensorbased MPPT method. Thus, in this paper, along with the proposed adaptive technique, steady-state behavior and drift analysis for the voltage-sensor-based MPPT method have been addressed. In this paper, single-ended primary-inductance converter (SEPIC) converter is considered because it works as stepup/step-down converter [16], [29]; thereby, it will increase the range of operation of PV voltage. This topology has merits of noninverting output polarity, easy to drive switch, and low input current ripple [16]. This paper is organized as follows. Voltage-sensor-based MPPT with fixed step size and adaptive step size are first addressed, and then, steady-state two-level operation and drift analysis for a change in insolation are presented in Section II. Simulation and experimental results are given in Sections III and IV, respectively. Finally, conclusions are presented in Section V. II. VOLTAGE -S ENSOR -B ASED MPPT FOR SEPIC MPPT controller is essential to extract the maximum power from the PV module or array. If the load is directly connected to the PV module, then it is not possible to operate at peak power point due to impedance mismatch. A converter acts as an interface to operate at MPP by changing the duty cycle generated from the MPPT controller. A general block diagram of a PV system with MPPT controller is shown in Fig. 1. From the plotted P −D characteristics shown in Fig. 2, it can be observed that the slope of the curve dP/dD = 0 at MPP, dP/dD > 0 to the left of MPP, and dP/dD < 0 to the right of MPP. Thus, the voltage-sensor-based MPPT algorithm for SEPIC converter is developed by using the P −D characteris-

Fig. 2. Normalized variation of P , P ∗ , Q with duty cycle for SEPIC.

tics obtained with the PV module. The objective function to implement this algorithm for SEPIC converter has been derived for both resistive and battery load as follows. A. Case 1: For Resistive Load Using input and output voltage relation for SEPIC (i.e., Vo = (D/(1 − D))VPV ), the efficiency of the converter can be expressed as ⎫ Vo Io η = VPV ⎪ ⎬ IPV 2 2 Req (1) Req Vo Io Vo D η = V 2 = VPV RL = 1−D RL ⎪ ⎭ PV Req

where VPV and IPV are the PV voltage and current, respectively. The equivalent input resistance Req of the converter can be obtained from (1) as follows:

1−D 2 Req = η RL . (2) D By using (2), the output power from the PV module, which is input power to the converter, is given by

2 V2 D V2 P = PV = PV . (3) Req ηRL 1 − D Fig. 2 shows that both P and square root of power (P ∗ ) have the maximum value at the same duty cycle (D). By considering the square root of power (P ∗ ) to obtain an objective function for tracking the maximum power, the following equation can be obtained:

√ D VPV P∗ = P = √ . (4) ηRL 1 − D At MPP from Fig. 2, the slope of the P ∗ curve is zero (i.e., dP ∗ /dD = 0), and it can be evaluated as

dP ∗ 1 1 D dVPV √ = VPV + = 0 (5) 2 dD (1 − D) 1 − D dD ηRL

VPV dD + D(1 − D)dVPV 1 dP ∗ √ = = 0. (6) dD (1 − D)2 dD ηRL By evaluating dP ∗ /dD using (6) at MPP, the objective function Q can be obtained as follows: ⎧ ⎪ ⎨= 0, at MPP Q = D(1−D)dVPV +VPV dD > 0, on left of MPP (7) ⎪ ⎩ < 0, on right of MPP.

KILLI AND SAMANTA: MPPT FOR PHOTOVOLTAIC SYSTEMS WITH SEPIC CONVERTER

Hence, depending on the sign of Q, the MPPT algorithm decides whether to increase or decrease the duty cycle, and the corresponding Q−D characteristics are shown in Fig. 2. B. Case 2: For Battery as a Load P = VPV IPV , and hence, dP/dD can be expressed by dVPV dIPV dP = IPV + VPV . (8) dD dD dD For SEPIC converter, IPV and battery current (Ibat ) can be expressed by (9), and hence, dIPV /dD can be given by (10), i.e., D IPV = Ibat (9) 1−D 1 dIPV = Ibat . (10) dD (1 − D)2 By substituting (9) and (10) in (8) D(1 − D)dVPV + VPV dD dP = Ibat . dD (1 − D)2 dD

(11)

At MPP, dP/dD = 0, as shown in Fig. 2, and hence, by evaluating (11) at MPP, the objective function (Q) for tracking the peak power with the battery load can be obtained as same as (7). Thus, the voltage-sensor-based MPPT method is valid for both resistive and battery load. The objective function Q for tracking the peak power in case of a PV system with boost converter can be obtained by evaluating the corresponding Req and dP ∗ /dD as follows: ⎧ ⎪ ⎨= 0, at MPP (12) Q = (1 − D)dVPV + VPV dD > 0, on left of MPP ⎪ ⎩ < 0, on right of MPP. From (7) and (12), it can be concluded that the objective function (Q) will vary depending on the dc–dc converter used with the PV system. Thus, the objective function (Q) for tracking the MPP depends on the converter topology for voltage-sensorbased MPPT method, whereas the widely accepted methods such as P&O and IncCond are independent of the converter. The two key parameters in any MPPT algorithm are perturbation time and perturbation step size, and the selection criteria for these two parameters are described as follows. 1) Selecting Proper Perturbation Time (Ta ): For a step change in duty cycle, the perturbation time should be greater than the settling time of the system [19]. Different values of ΔD will result in different values of settling time. The perturbation time is chosen such that it should be greater than the settling time for a maximum step ΔDmax change in duty cycle. 2) Selecting Proper Perturbation Step Size (ΔD): The perturbation step size should be chosen by considering dynamic and steady-state performance. The maximum value of step size (ΔDmax ) improves the dynamic performance, whereas the minimum value of step size (ΔDmin ) results in lower oscillations around the MPP, which, in turn, improves the steady-state performance. The step size (ΔDmin ) should be chosen based on the tracking accuracy in steady state [26] and the analog-todigital (ADC) resolution [30] of the microcontroller used in the system. The standard procedure for choosing ΔDmin value is

7611

that the minimum voltage change due to perturbation of D by ΔDmin should be more than the ADC resolution. For an N -bit ADC with a maximum value of the ADC channel as VADC , it should satisfy the following condition: VADC (13) [V (D + ΔDmin ) − V (D)] ∗ S ≥ N 2 where S is the scaling factor, which is equal to R2 /(R1 + R2 ) in this case. Thus, ΔDmin will vary with different PV panel characteristics and different ADCs. In this case, ΔDmin is chosen as 0.5% for a 10-bit ADC with VADC of 5 V, which satisfies (13). C. Adaptive Voltage-Sensor-Based MPPT With Variable Scaling Factor In this paper, an adaptive voltage-sensor-based MPPT with variable scaling factor is proposed to reduce the tracking time and power loss in steady state. The present and previous iteration values of PV voltage and duty cycle of the converter are denoted by VPV (k), VPV (k − 1), D(k), and D(k − 1), respectively. The changes in the voltage and duty cycle from the present iteration to the next iteration are defined as follows: dVPV = VPV (k) − VPV (k − 1) dD = D(k) − D(k − 1).

(14) (15)

The location of the operating point is decided by evaluating Q and depending on the sign of Q; the duty cycle is incremented or decremented by ΔD as given in (16). If Q is positive, then the duty cycle is incremented by ΔD, and if Q is negative, then the duty cycle is decremented by ΔD. As ΔD is directly used in adjusting the duty cycle, the controller is simple and easy to implement with a microcontroller. Thus, D(k + 1) = D(k) ± ΔD.

(16)

Variations of Q using the experimental data in start-up case for a change in insolation from 0 to 270 W/m2 and for a change in insolation from 270 to 480 W/m2 are plotted in Fig. 3(a) and (b), respectively. Fig. 3 shows that the value of Q is large in start-up and during insolation change, whereas it is small in the steady state. Thus, a fixed scaling factor cannot satisfy the requirement of MPPT controller in different conditions. Hence, in this proposed algorithm, two different scaling factors M1 and M2 are considered to optimally vary the perturbation step size ΔD, which has been defined as a linear function of Q by ΔD = Mi Q.

(17)

The scaling factor Mi (i = 1, 2) plays a significant role in an adaptive MPPT method; therefore, it should be chosen judiciously to increase the peak power tracking efficiency. The scaling factor M1 is chosen to reduce the tracking time in startup and for a large change in insolation. The scaling factor M2 is chosen to reduce the power loss in the steady state. Thus, the proposed adaptive MPPT method improves both the transient and steady-state performance. The scaling factor either M1 or M2 is chosen to generate ΔD depending on the value of Q with respect to a predefined threshold value of the objective function, i.e., Qth , as shown

7612


Fig. 4. Movement of the operating point on the Q−D characteristics along with P −V characteristics.

Fig. 3. Variation of Q (a) from start-up to steady state and (b) for a step change in solar insolation.

in the pseudocode of the algorithm. By considering an upper limit (ΔDmax ) of 10% and a lower limit (ΔDmin ) of 0.5% to perturbation step size (ΔD), the scaling factors M1 and M2 should obey (18) and (19), respectively, in order to guarantee the convergence of the MPPT algorithm. The value of ΔD will vary between ΔDmin and ΔDmax , as given in (20). Thus, M1 Q ≤ ΔDmax M2 Q ≥ ΔDmin ⎧ ⎪ ⎨ΔDmax , if ΔD > ΔDmax ΔD = ΔD, if ΔDmax ≤ ΔD ≤ ΔDmin ⎪ ⎩ ΔDmin , if ΔD < ΔDmin .

(18) (19) (20)

The pseudocode for the proposed algorithm is given in the following. Initialize VPV (k − 1) and D(k − 1) Initialize M1 , M2 , and Qth Loop: Sample and average VPV (k) Calculate Q If(|Q| ≥ Qth ) ΔD = M1 |Q| OR If(|Q| < Qth ) ΔD = M2 |Q| If(Q > 0) D(k + 1) = D(k) + ΔD OR If(Q < 0) D(k + 1) = D(k) − ΔD ELSE No Change D(k + 1) = D(k) GOTO Loop

D. Steady-State Analysis The movement of the operating point on Q−D characteristics and the corresponding point on P −V characteristics is shown in Fig. 4. Assume that the operating point during a As Q > 0 at point , a (k − 3)Ta time interval is at point . the algorithm increases the duty cycle, and hence, the operating b during (k − 2)Ta time interval. At point moves to point b the algorithm again increases the duty cycle as Q > 0, point , c Similarly at point , c and the operating point moves to point . the algorithm increases the duty cycle because Q > 0, and d during kTa time hence, the operating point moves to point d as Q < 0, the algorithm decreases the duty interval. At point , c cycle, and hence, the operating point moves back to point . c as Q > 0, the algorithm makes the operating Again at point , d by increasing the duty cycle. Thus, in point to move to point steady state, the operating point moves in two levels, resulting in power loss reduction compared with P&O and IncCond because, in case of P&O [19] and IncCond [31], the operating point moves in three levels, as shown in Fig. 5. E. Steady-State Power Loss Evaluation The two-level operation of the voltage-sensor-based MPPT algorithm reduces the voltage oscillations around the MPP, resulting in power loss reduction compared with three-level MPPT algorithms such as P&O and IncCond. The steady-state power loss calculation in case of P&O has been addressed in [32] and [33]. The power losses Pr due to oscillations in comparison with the available maximum power (Pmp ) are expressed in [32] as

2

(ΔVPV )RMS Pr VMPP = 1+ (21) Pmp VMPP 2avt where (ΔVPV )RMS is the RMS value of steady-state voltage oscillation, VMPP is the PV module voltage at MPP, a is the diode ideal factor, and Vt is the thermal voltage of the PV


7613

The RMS value of the voltage oscillations for P&O can be obtained by T1 √ 3 2 . (26) (ΔVPV (t)) dt = ΔV (ΔVPV )RMSP&O = 2 0

The relative power loss with P&O MPPT algorithm can be obtained by substituting (26) into (21), i.e.,

Pr VMPP 3 ΔV 2 1+ . (27) ≈ Pmp P&O 4 VMPP 2avt For the considered PV module with VMPP = 15.5 V at an insolation of 270 W/m2 , the voltage oscillations around the MPP are ΔV = 0.5 V, as shown in Fig. 5. The relative power loss for the voltage-sensor-based MPPT and P&O can be obtained by evaluating (24) and (27), respectively, for the specific parameters used in this work as follows:

Pr ≈ 0.19% (28) Pmp VS

Pr ≈ 0.57%. (29) Pmp P&O The tracking efficiency of MPPT, i.e., ηMPPT , algorithm can be obtained as

Pr ηMPPT = 1 − × 100%. (30) PMPP Fig. 5. Steady-state voltage oscillations around the MPP (a) for voltage-sensor-based MPPT and (b) for P&O MPPT.

module. The voltage oscillations around the MPP, as shown in Fig. 5(a), with the voltage-sensor-based MPPT are given by 1 ΔV, if 0 < t ≤ T2 ΔVPV (t) = 21 (22) T 2 ΔV, if 2 < t ≤ T. The RMS value of the voltage oscillations for voltage-sensorbased MPPT algorithm is given by T 1 (23) (ΔVPV )RMSVS = (ΔVPV (t))2 dt = ΔV . 2

The tracking efficiency of voltage-sensor-based MPPT, i.e., ηVS and P&O, i.e., ηP&O , by considering only the power loss due to voltage oscillations around the MPP in steady state can be evaluated by substituting (28) and (29) into (30) as follows: ηVS = 99.81%

(31)

ηP&O = 99.43%.

(32)

The relative power loss for voltage-sensor-based MPPT is only 0.19% of available maximum power, which is one third of the power loss for other MPPT methods with three-level steadystate operation (for example, P&O). Over a 25-year life cycle of a PV setup, a remarkable energy gain can be achieved with this method.

0

Substituting (23) into (21) results in relative power loss with the voltage-sensor-based MPPT in steady state as follows: 2

Pr ΔV 1 VMPP ≈ 1+ . (24) Pmp VS 4 VMPP 2avt From Fig. 5(b), the voltage oscillations around the MPP in steady state in case of three-level MPPT algorithms such as P&O and IncCond are given by ⎧3 T1 ⎪ 2 ΔV, if 0 < t ≤ 4 ⎪ ⎪ ⎨ 1 ΔV, if T1 < t ≤ T1 4 2 (25) ΔVPV (t) = 21 3T1 T ⎪ ΔV, if < t ≤ ⎪ 2 2 4 ⎪ ⎩1 3T1 2 ΔV, if 4 < t ≤ T1 .

F. Drift Analysis The movement of the operating point in a wrong direction for a change in insolation is called drift, and this effect is severe in case of rapid change in insolation [19]–[22]. The drift problem occurs in case of change in insolation with P&O and Inccond methods, and it is well addressed in the literature [18]–[20], [22], but the drift analysis for voltage-sensor-based MPPT does not exist in the literature. The drift analysis with this method can be examined by evaluating the change in operating voltage VPV and the objective function Q for a change in insolation. The relation between IPV and VPV corresponding to the present operating point on the I−V characteristics of the PV

7614


Fig. 7. Simulation and experimental P −V characteristics of the PV module for two different insolation levels.

Suppose there is an insolation change from G1 to G2 in between the perturbation time while operating with a certain duty cycle. Then the change in operating voltage due to the insolation change can be obtained by using (35) as follows: VPV |G2 =

Isc |G2 Irs D2 ηRL (1−D)2 + aVt

.

(36)

From (35) and (36), it can be noticed that, for an increase in insolation, VPV increases as Isc increases, and the change in VPV due to insolation change can be obtained as follows: ΔVPVG =

Isc |G2 − Isc |G1 D2 ηRL (1−D)2

+

Irs aVt

.

(37)

By substituting (36) and (37) into (7), the objective function Q for a change in insolation while operating with a certain duty cycle D can be obtained as follows: Q = D(1 − D) Fig. 6. Drift analysis with Q−D and P −V characteristics (a) for a step increase in insolation and (b) for a step decrease in insolation.

module with SEPIC converter can be expressed in terms of slope of the load line as IPV =

D2 VPV . ηRL (1 − D)2

(33)

From the ideal single-diode model of the PV module, the current and voltage relationship can be described as follows [1]: VPV (34) IPV = Isc − Irs e aVt − 1 where Isc is the short-circuit current, and Irs is the reverse saturation current. By substituting (33) into (34) and by considering Taylor’s series expansion of the exponential term up to first order, VPV can be expressed in terms of Isc for an insolation G1 as follows: VPV |G1 =


.

(35)

Isc |G2 − Isc |G1 D2 ηRL (1−D)2

+

Irs aVt

+


dD.

(38) From (38), it can be verified that Q is positive for an increase in insolation and is negative for a decrease in insolation as Isc is directly proportional to the insolation level. As the MPPT algorithm takes the decision based on the sign of Q, hence, for an increase in insolation, the algorithm increases the duty cycle, i.e., moving toward the MPP, and for a decrease in insolation, the algorithm decreases the duty cycle, i.e., moving toward the MPP. Thus, the voltage-sensor-based MPPT method is free from drift for both increase and decrease in insolation conditions. Pictorial representation of the drift analysis is shown in Fig. 6. As shown in Fig. 6(a), assume that there is an increase in d then the operating point insolation while operating at point ; e on the Q−D or the P −V curve will be settled to a new point corresponding to the increased insolation. Now, the algorithm e takes a decision to increase the duty cycle as Q > 0 at point , f and thereby, the operating point moves closer to MPP (point ). Thus, the voltage-sensor-based MPPT method is free from drift in case of increase in insolation. Similarly, for a decrease in d as shown in Fig. 6(b), the insolation while operating at point ,


7615

Fig. 9. Circuit model of the developed PV system.

Fig. 10. Experimental setup of the developed PV system.

Fig. 8. Tracking waveforms with fixed-step-size and adaptive-step-size methods for resistive load. (a) Voltage. (b) Power. (c) Duty cycle.

e on the Q−D operating point will be settled to a new point or the P −V curve corresponding to the decreased insolation. e the algorithm decreases the duty cycle, As Q < 0 at point , f and hence, the operating point moves closer to MPP (point ). Thus, the voltage-sensor-based MPPT algorithm is free from drift for both increase and decrease in insolation. III. S IMULATION R ESULTS For simulation studies, the PV module is modeled by using five-parameter single-diode circuit model [1], [34], and those five parameters at nominal conditions are chosen as IPV,n = 1.9 A, I0 = 23.089.10−8 A, a = 1.3, Rs = 0.32 Ω, and Rp = 120 Ω. The components for the designed SEPIC converter

used in simulation and the experimental setup are chosen as L1 = 180 μH, L2 = 180 μH, C1 = 47 μF, C2 = 220 μF, Cin = 440 μF, fs = 50 kHz, and RL = 25 Ω. The proposed MPPT algorithm has been tested for a step change in insolation level from 270 to 480 W/m2 at 0.1 s. The perturbation time and the perturbation step size for the fixed-step-size method are chosen as 20 ms and 0.5%, respectively. The P −V characteristics of the simulated PV module are shown in Fig. 7. From the simulated P −V characteristics, it can be noticed that the MPP voltages are 15.13 and 16.16 V, corresponding to the insolation level of 270 and 480 W/m2 , respectively. The tracking waveforms with the fixed-step-size method and with the proposed adaptive method are shown in Fig. 8, and it can be visualized that both methods are efficiently tracking the corresponding MPP voltage, but the tracking time is larger in the case of the fixed-step-size method. Fig. 8(a) shows that the tracking time with the fixed-step-size method in start-up case for G = 270 W/m2 is T1 = 600 ms, and it has been reduced to 130 ms with the proposed adaptive method. Similarly for an increase in insolation from 270 to 480 W/m2 , the tracking time has been reduced from 290 to 160 ms with the adaptive method compared with the fixed-step-size method. From Fig. 8(c), it can be noticed that the duty cycle is varying between two values in steady state, resulting in power loss reduction. From the results shown in Fig. 8, it can be concluded that both dynamic

7616


Fig. 11. Tracking waveforms with fixed-step-size MPPT for resistive load (a) start-up for (G, T ) = (270 W/m2 , 43 ◦ C) and (b) for a change in solar insolation from (G, T ) = (270 W/m2 , 43 ◦ C) to (480 W/m2 , 48 ◦ C).

and steady-state performance are improved by the proposed adaptive MPPT algorithm. IV. E XPERIMENTAL VALIDATION To substantiate the functionality and performance of the proposed system, a prototype of the SEPIC converter and control circuit has been developed. The SEPIC converter has been implemented using the designed parameters presented in Section III. The ARDUINO ATMEGA 2560 has been used to implement the MPPT algorithm and to provide the pulsewidth modulation control signal to the SEPIC converter. The circuit model of the proposed system and the experimental setup are shown in Figs. 9 and 10, respectively. Voltage measurement is required for implementation of the MPPT algorithm, and the microcontroller board cannot tolerate more than 5 V. Thus, the voltage is measured using the voltage divider circuit with resistances R1 and R2 of values 10 and 1 kΩ, respectively. The measurement noise with a sense resistor is less compared with a Hall-effect transducer. However, in case of low signal-to-noise ratio, the effect of noise can be mitigated by reducing the sampling frequency or by increasing the perturbation step size [35] and by using low-pass filter

Fig. 12. Tracking waveforms with adaptive-step-size MPPT for resistive load (a) start-up for (G, T ) = (270 W/m2 , 43 ◦ C) and (b) for a change in solar insolation from (G, T ) = (270 W/m2 , 43 ◦ C) to (480 W/m2 , 48 ◦ C) and vice versa.

[36]. The PV module chosen for the experimental setup having model number ELDORA 40-P is shown in Fig. 10, and the experiment is carried out using the artificial insolation with the help of halogen and incandescent lamps. The voltage-sensor-based MPPT technique with fixed step size and the proposed adaptive-step-size method are tested for a step change in insolation from 270 to 480 W/m2 . The measured P −V characteristics of the considered PV module at (G, T ) = (270 W/m2 , 43 ◦ C) and at (G, T ) = (480 W/m2 , 48 ◦ C) are shown in Fig. 7. From the experimental data, it can be noticed that the voltages corresponding to MPP for the aforementioned insolation and temperature conditions are 15.5 and 16.5 V, respectively. The tracking waveforms with a fixed step size of 0.5% at (G, T ) = (270 W/m2 , 43 ◦ C) is shown in Fig. 11(a), and it can be observed that the tracking time for start-up is T1 = 750 ms. Fig. 11(b) shows the tracking waveforms for a change in insolation and temperature from (G, T ) = (270 W/m2 , 43 ◦ C) to (G, T ) = (480 W/m2 , 48 ◦ C) with a fixed step size of 0.5%, and it can be noticed that it takes a tracking time of T2 = 550 ms. The tracking waveforms with the adaptive MPPT method corresponding to start-up and change in insolation level are shown in Fig. 12(a) and (b),


7617

Fig. 13. Tracking waveforms with adaptive-step-size MPPT for battery load (a) start-up for (G, T ) = (270 W/m2 , 43 ◦ C) and (b) for a change in solar insolation from (G, T ) = (270 W/m2 , 43 ◦ C) to (480 W/m2 , 48 ◦ C) and vice versa.

respectively. Fig. 12(a) and (b) shows that the tracking time T1 is reduced to 250 ms from 750 ms and T2 is reduced to 150 ms from 550 ms with the proposed method compared with the fixed-step-size method. The tracking time with the P&O and Inccond methods is greater compared with the voltage-sensorbased MPPT due to processing of another measurement signal (current) and evaluation of extra conditions in those algorithms. Thus, the proposed adaptive MPPT method is efficient in terms of reducing tracking time and power loss in steady state. The experiment is also performed by considering a lead-acid battery (12 V, 1.5 Ah) as a load, and the corresponding tracking waveforms for start-up and change in insolation level are shown in Fig. 13(a) and (b), respectively. Based on the experimental results shown in Fig. 13, it can be observed that the MPPT algorithm is effectively tracking the peak power with the battery load. The SEPIC converter always operates in step-down mode for this specific battery as Vb < VMPP (i.e., DMPP < 0.5 for SEPIC), where Vb is the battery voltage. On the other hand, with resistive load, the SEPIC converter will operate either in step-down or step-up mode depending on the impedance matching. Hence, the variation of duty cycle is less for a change in insolation, which results in the reduction of tracking time

Fig. 14. Tracking waveforms with adaptive voltage-sensor-based MPPT method with Ta = 1s for resistive load. (a) Steady-state two-level operation for (G, T ) = (270 W/m2 , 43 ◦ C). (b) Drift analysis for an increase in insolation from G = 270 W/m2 to G = 480 W/m2 . (c) Drift analysis for a decrease in insolation from G = 480 W/m2 to G = 270 W/m2 .

with the battery load as compared with the PV system with resistive load, as shown in Figs. 12 and 13. For precise observation of steady-state behavior and drift phenomena of the proposed adaptive method, experiment is carried out with Ta = 1s, and the corresponding experimental results are shown in Fig. 14. Fig. 14(a) shows that the

7618


TABLE I C OMPARISON OF VOLTAGE -S ENSOR -B ASED MPP W ITH THE P&O AND I NC C OND M ETHODS

voltage-sensor-based MPP method is exhibiting two-level operation in steady state, which will effectively reduce the power loss compared with P&O and IncCond, as mentioned in Section II. The drift analysis is carried out for a step increase and a step decrease in insolation from (G, T ) = (270 W/m2 , 43 ◦ C) to (G, T ) = (480 W/m2 , 48 ◦ C) and from (G, T ) = (480 W/m2 , 48 ◦ C) to (G, T ) = (270 W/m2 , 43 ◦ C) in between the perturbation time interval, respectively. As shown in Fig. 14(b), due to increase in insolation, the operating voltage is increased to a voltage level more than the MPP voltage, and in the next cycle, the voltage is reduced (power increased) as Q > 0. Similarly for decrease in insolation, as shown in Fig. 14(c), the operating voltage is reduced to a voltage level less than the MPP voltage, and in the next cycle, the voltage is increased (power is increased) as Q < 0. Thus, the voltage-sensor-based MPP method is free from drift for both increase and decrease in insolation. The noise in the current measurement shown in experimental results is due to inherent noisy nature of Hall-effect transducer (LEM LA-55P) [37] and to high switching frequency of the converter. The comparison of the voltage-sensor-based MPPT method with the two widely used methods P&O and IncCond is presented in Table I.

V. C ONCLUSION In this paper, an adaptive voltage-sensor-based MPPT algorithm with variable scaling factor by considering direct duty cycle control method for SEPIC converter has been implemented. The proposed system is designed, and the functionality of MPPT control has been proved. The simulation and experimental results prove that the proposed system is able to track the maximum power from the PV module; moreover, the steady-state two-level operation and the drift-free phenomena are the merits of this tracking algorithm. Hence, this method improves the efficiency of the PV system and reduces power loss in steady state. From the results obtained, it is noticed that, with a well-designed system, including a proper converter and an efficient MPPT algorithm, the MPPT can be developed with less complexity and reduced cost.

R EFERENCES [1] M. G. Villalva, J. R. Gazoli, and E. R. Filho, “Comprehensive approach to modeling and simulation of photovoltaic arrays,” IEEE Trans. Power Electron., vol. 24, no. 5, pp. 1198–1208, May 2009. [2] J. H. R. Enslin, M. S. Wolf, D. B. Snyman, and W. Swiegers, “Integrated photovoltaic maximum power point tracking converter,” IEEE Trans. Ind. Electron., vol. 44, no. 6, pp. 769–773, Dec. 1997. [3] M. A. S. Masoum, H. Dehbonei, and E. F. Fuchs, “Theoretical and experimental analyses of photovoltaic systems with voltage and current-based maximum power-point tracking,” IEEE Trans. Energy Convers., vol. 17, no. 4, pp. 514–522, Dec. 2002. [4] W. Xiao and W. G. Dunford, “A modified adaptive hill climbing MPPT method for photovoltaic power systems,” in Proc. IEEE PESC, 2004, pp. 1957–1963. [5] D. Sera, R. Teodorescu, J. Hantschel, and M. Knoll, “Optimized maximum power point tracker for fast-changing environmental conditions,” IEEE Trans. Ind. Electron., vol. 55, no. 7, pp. 2629–2637, Jul. 2008. [6] K. L. Lian, J. H. Jhang, and I. S. Tian, “A maximum power point tracking method based on perturb-and-observe combined with particle swarm optimization,” IEEE J. Photovoltaics, vol. 4, no. 2, pp. 626–633, Mar. 2014. [7] F. Liu, S. Duan, B. Liu, and Y. Kang, “A variable step size INC MPPT method for PV systems,” IEEE Trans. Ind. Electron., vol. 55, no. 7, pp. 2622–2628, Jul. 2008. [8] A. Safari and S. Mekhilef, “Simulation and hardware implementation of incremental conductance MPPT with direct control method using Cuk converter,” IEEE Trans. Ind. Electron., vol. 58, no. 4, pp. 1154–1161, Apr. 2011. [9] K. S. Tey and S. Mekhilef, “Modified incremental conductance algorithm for photovoltaic system under partial shading conditions and load variation,” IEEE Trans. Ind. Electron., vol. 61, no. 10, pp. 5384–5392, Oct. 2014. [10] Q. Mei, M. Shan, L. Liu, and J. M. Guerrero, “A novel improved variable step-size incremental-resistance MPPT method for PV systems,” IEEE Trans. Ind. Electron., vol. 58, no. 6, pp. 2427–2434, Jun. 2011. [11] T. Esram, J. W. Kimball, P. T. Krein, P. L. Chapman, and P. Midya, “Dynamic maximum power point tracking of photovoltaic arrays using ripple correlation control,” IEEE Trans. Power Electron., vol. 21, no. 5, pp. 1282–1291, Sep. 2006. [12] T. Wu, C. Chang, and Y. Chen, “A fuzzy-logic-controlled single-stage converter for PV-powered lighting system applications,” IEEE Trans. Ind. Electron., vol. 47, no. 2, pp. 287–296, Apr. 2000. [13] Syafaruddin, E. Karatepe, and T. Hiyama, “Artificial neural network-polar coordinated fuzzy controller based maximum power point tracking control under partially shaded conditions,” IET Renew. Power Gener., vol. 3, no. 2, pp. 239–253, Jul. 2009. [14] K. Ishaque and Z. Salam, “A deterministic particle swarm optimization maximum power point tracker for photovoltaic system under partial shading condition,” IEEE Trans. Ind. Electron., vol. 60, no. 8, pp. 3195–3206, Aug. 2013. [15] E. Bianconi et al., “A fast current-based MPPT technique employing sliding mode control,” IEEE Trans. Ind. Electron., vol. 60, no. 3, pp. 1168–1178, Mar. 2013. [16] E. Mamarelis, G. Petrone, and G. Spagnuolo, “Design of a sliding mode controlled SEPIC for PV MPPT applications,” IEEE Trans. Ind. Electron., vol. 61, no. 7, pp. 3387–3398, Jul. 2014. [17] T. Esram and P. L. Chapman, “Comparison of photovoltaic array maximum power point tracking techniques,” IEEE Trans. Energy Convers., vol. 22, no. 2, pp. 439–449, Jun. 2007. [18] D. P. Hohm and M. E. Ropp, “Comparative study of maximum power point tracking algorithms,” Prog. Photovoltaics Res. Appl., vol. 11, no. 1, pp. 47–62, Apr. 2003. [19] N. Femia, G. Petrone, G. Spagnuolo, and M. Vitelli, “Optimization of perturb and observe maximum power point tracking method,” IEEE Trans. Power Electron., vol. 20, no. 4, pp. 963–973, Jul. 2005. [20] M. A. Algendy, B. Zahawi, and D. J. Atkinson, “Assessment of perturb and observe MPPT algorithm implementation techniques for PV pumping applications,” IEEE Trans. Sustain. Energy, vol. 3, no. 1, pp. 21–33, Jan. 2012. [21] K. H. Hussein, I. Muta, T. Hoshino, and M. Osakada, “Maximum photovoltaic power tracking: An algorithm for rapidly changing atmospheric conditions,” Proc. Inst. Elect. Eng.—Gener., Transmiss. Distrib., vol. 142, no. 1, pp. 59–64, Jan. 1995. [22] D. Sera, L. Mathe, T. Kerekes, S. V. Spataru, and R. Teodorescu, “On the perturb-and-observe and incremental conductance MPPT methods for PV systems,” IEEE J. Photovoltaics, vol. 3, no. 3, pp. 1070–1078, Jul. 2013.


[23] V. Salas, E. Olias, A. Barrado, and A. Lazaro, “Review of the maximum power point tracking algorithms for stand-alone photovoltaic systems,” Solar Energy Mater. Solar Cells, vol. 90, no. 11, pp. 1555–1578, Jul. 2006. [24] M. A. G. de Brito, L. Galotto, L. P. Sampaio, G. A. E. Melo, and C. A. Canesin, “Evaluation of the main MPPT techniques for photovoltaic applications,” IEEE Trans. Ind. Electron., vol. 60, no. 3, pp. 1156–1167, Mar. 2013. [25] L. Piegari and R. Rizzo, “Adaptive perturb and observe algorithm for photovoltaic maximum power point tracking,” IET Renew. Power Gener., vol. 4, no. 4, pp. 317–328, Jul. 2010. [26] N. Femia, D. Granozio, G. Petrone, G. Spagnuolo, and M. Vitelli, “Predictive and adaptive mppt perturb and observe method,” IEEE Trans. Aerosp. Electron. Syst., vol. 43, no. 3, pp. 934–950, Jul. 2007. [27] M. Veerachary, T. Senjyu, and K. Uezato, “Voltage-based maximum power point tracking control of PV system,” IEEE Trans. Aerosp. Electron. Syst., vol. 38, no. 1, pp. 262–270, Jan. 2002. [28] N. Dasgupta, A. Pandey, and A. K. Mukerjee, “Voltage-sensing-based photovoltaic MPPT with improved tracking and drift avoidance capabilities,” Solar Energy Mater. Solar Cells, vol. 92, no. 12, pp. 1552–1558, Jun. 2008. [29] S. J. Chiang and H. J. Shieh, “Modeling and control of PV charger system with SEPIC converter,” IEEE Trans. Ind. Electron., vol. 56, no. 11, pp. 4344–4353, Nov. 2009. [30] Y. Jiang, J. A. A. Qahouq, and T. A. Haskew, “Adaptive step size with adaptive-perturbation-frequency digital MPPT controller for a singlesensor photovoltaic solar system,” IEEE Trans. Power Electron., vol. 28, no. 7, pp. 3195–3205, Jul. 2013. [31] M. A. Algendy, B. Zahawi, and D. J. Atkinson, “Assessment of the incremental conductance maximum power point tracking algorithm,” IEEE Trans. Sustain. Energy, vol. 4, no. 1, pp. 108–117, Jan. 2013. [32] C. R. Sullivan, J. J. Awerbuch, and A. M. Latham, “Decrease in photovoltaic power output from ripple: Simple general calculation and the effect of partial shading,” IEEE Trans. Power Electron., vol. 28, no. 2, pp. 740–747, Feb. 2013. [33] F. Paz and M. Ordonez, “Zero oscillation and irradiance slope tracking for photovoltaic MPPT,” IEEE Trans. Ind. Electron., vol. 61, no. 11, pp. 6138–6147, Nov. 2014. [34] Y. A. Mahmoud, W. Xiao, and H. H. Zeineldin, “A parameterization approach for enhancing PV model accuracy,” IEEE Trans. Ind. Electron., vol. 60, no. 12, pp. 5708–5716, Dec. 2013.

7619

[35] A. M. Latham, R. Pilawa-Podgurski, K. M. Odame, and C. R. Sullivan, “Analysis and optimization of maximum power point tracking algorithms in the presence of noise,” IEEE Trans. Power Electron., vol. 28, no. 7, pp. 3479–3494, Jul. 2013. [36] H. Al-Atrash, I. Batarseh, and K. Rustom, “Effect of measurement noise and bias on hill-climbing MPPT algorithms,” IEEE Trans. Aerosp. Electron. Syst., vol. 46, no. 2, pp. 745–760, Apr. 2010. [37] C. Cavallaro, F. Chimento, S. Musumeci, and A. Raciti, “A voltage sensing approach for a maximum power tracking in integrated photovoltaic applications,” in Proc. Int. Conf. Clean Elect. Power, 2009, pp. 691–698. Muralidhar Killi received the B.Tech. degree in electronics and communication engineering from Velagapudi Ramakrishna Siddhartha Engineering College, Vijayawada, India, in 2008 and the M.Tech. degree in electrical engineering from the National Institute of Technology, Rourkela, India, in 2012, where he is currently working toward the Ph.D. degree in electrical engineering. His areas of research interest include modeling, analysis, and control of photovoltaic systems, and power electronic converter circuits.

Susovon Samanta (M’11) received the B.E. degree from the Regional Engineering College, Durgapur, India, in 1998, the M.E. degree from Jadavpur University, Kolkata, India, in 2003, and the Ph.D. degree from the Indian Institute of Technology, Kharagpur, Kharagpur, India, in 2012, all in electrical engineering. From 2003 to 2004, he was a Lecturer with Birla Institute of Technology, Mesra, India. In 2009, he joined the National Institute of Technology, Rourkela, India, where he is currently an Assistant Professor. His research interests are in the analysis, modeling, and control of power converters used in various areas such as photovoltaics, electric vehicles, and space applications.